Books or Academic Papers?

These days, maybe because it has been gloomy outside, I have a lingering mood to do research related activities indoor. Just kidding, I have been trying to accumulate my knowledge in the field for quite a while. My favorite way to do this is to read books by scholar. But they are in general pretty thick and I guess predictably heavy to read, so I must take my time to read them if I want to understand them and record them in my brain. And since my schedule is very variant, I never know, sometimes I might have a long free time when I can read my book, so I always try to have the book within my reach in my bag or more frequently, on my hand.

Retrieved from Google on 9 May 2018. All rights reserved.

So one day, when I was on my way back from work, I met my ex-college professor. I was holding one of those books on my hand and that caught my professor’s eyes. He asked me if I am reading that book and if I like it or not. I told him, I am in the middle of reading it but so far I enjoyed it a lot and I asked him if he had read it too. His answer was what I has half expected, “No, I only read his academic papers. Do you read the paper?” Well, I said no (I did for some but only the abstract so I think that does not count right?) Upon hearing my answer, his expression changed to what I think an underestimating expression. It might be that he didn’t mean it that way, but since he was also a professor who told me that women cannot pursue economics PHD, I can’t help to be biased. Anyway in this post I would like to share the reason why I prefer to read the book instead of the academic papers.

Need not to be said, the book content was a lot easier in the sense of they are using layman’s language, very little jargon and less technical. It is also clear that it is less boring to read the book and less effort in finding the goods. Seems that reading academic papers gave out the signal of effort and academic like approach and book is for every other people. But really, reading the book can be more enriching and inspiring than academic papers. Academic papers give us information about one particular study, from how they did that, what had been done so far in the topic, why is that significant (according to the researcher). Sometimes, the researcher had done several papers on the same topic and mentioned that in their introduction or reference. But more often, they have a range of research interests which might not necessarily related one another. We might be able to find all their paper in their profile but reading one paper to another creates a gap because it is unclear how this particular researcher think, why is he/she interested in these very different research areas. The thinking process and idea generation steps, which absent from the body of academic paper, are offered in the published book. The books, despite chunked into a number of chapters, provide the story behind one’s research journey. It introduce a more personal approach in doing research. Ideas rarely fell from the tree (except for Newton’s gravity theory), they have history. Since my research interest lies in human irrationality, I prefer reading books by behavioral economist and psychologist. I learned that researching in this area may start from observing our own behavior and see if the misconduct prevalent within the society (or is it just me who’s weird).

Retrieved from Google on 9 May 2018. All rights reserved.

Another perk of reading the books is to gain implicit moral support. When we read through the academic paper, only the successes were shared. Little can be known about the hard path leading to the production of that paper. The possible failures and the length of time needed before that paper got published. Possible discrimination, bias and belittlement they might have faced. The importance of persistence, never-ending curiosity and belief as well as the love in learning. These are not given in academic papers.

So there are at least three benefits in reading books-written-by-researcher; one, it is easier to understand and it summarized all the important points (if you are skeptical, feel free to find the paper) a.k.a it is convenient. Two, it offers stories, let us understand the creative thinking and this might lead to inspiration for the readers to connect the dots. Lastly, it provide moral support and remind us that even the top prominent academics are still human and strong will is an important factor to get to the top.
Next time you met a professor, tell them you read academic papers if you want to increase your ‘I-am-so-intelligent’ signal, but behind your professor back, try to consider reading the books because there might be where you gain inspiration and lead you to great research idea.

If you are interested, here are some book recommendations:
– Misbehaving – Richard Thaler (Behavioral economist, Nobel winning economist 2017)
– The Upside of Irrationality – Dan Ariely (Psychologist and behavioral economist)
– Thinking, Fast and Slow – Daniel Kahneman (Psychologist, research interest in psychology of judgment and decision-making)


Coding Might be the Last Thing You Want to Do First – Step by Step (Part III)

I promise you this is the last part, I started to get tired of typing this too. But stay with me for the last step.

Step 6: Translate the result into a story
So after you run your regression, you will get tables and numbers maybe with stars generated by your statistical tools right. That means you have transformed your data into statistical information. But this result might be incomprehensible by non-statistics people, so the last step is to translate it to human language. This step is crucial for a classic reason: no matter what you find if no one understand it then it’s useless.
To do this, go back to your question, try to answer it with the information from the regression, but answer with tone like “discount is positively related to bread sales, 1% increase in discount resulted in XX% increase in bread sales and the result is significant in 5% significant level”, or something like that.


Retrieved from Google on 3 May 2018. All rights reserved.

I think I become lazy after writing 3 posts of one thousand plus words, but this part is really just this but super duper important.
I received some question of univariate and multivariate in SPSS. Apparently, they are called univariate when there is only one DV and multivariate if there is more than one DV. Remember here when I say univariate and multivariate is concerning the IV (Independent variable on the right hand side of the equation).

So yes, feel free to drop me question if you need help or if you spotted any mistake or any things that need clarification.
Thank you for reading and good luck!

Coding Might be the Last Thing You Want to Do First – Step by Step (Part II)

Continuation from Coding Might be the Last Thing You Want to Do First – Step by Step (Part I) and this post is all about technical things.

Step 4: Regression or just t-test?
This sounds technical, but don’t be intimidated right away. Before deciding which and how to do those regression and t-test, firstly take your question and variable list. For each question, ask how you think is better to answer it?
a. Comparing variables at it’s baseline value. Do you want to compare between two variables or one variable to a standard mean? Why do you compare? what information will you get after you compare, what if it is larger/smaller/equal? If the answer help you in answering the top question then you do t-test. There are of course different kind of t-test, so you need to go to your variable now and see, are they independent or not? then you can open your Google and ask “How to do (in)dependent t-test comparing (one or two) variables (if one to a number) in (stata/spss/r/whatever)” then voila it will give you the steps. i.e. do female buy more bread than male? this is independent t-test if the female and male are strangers and just random but will be dependent (or sometimes called paired) t-test if the female-male is for example wife and husband, so they are related.

Retrieved from Google on 3 May 2018. All rights reserved.

b. Finding correlation between variables. Do you want to know if one variable correlates to other i.e. if bread sales increase or decrease when there is a discount: variable sales and discount, see if they are correlated. If this what you want, opt for regression. Whenever you do regression, always remember the wise phrase: CORRELATION is NOT causality. Of course it is tempting to say: discount caused more/less sales of bread, but remember that regression shows correlation and not causality, there might be other event(s) that cause both independently to happen, in this case i.e. trend to make avocado toasts and storehouse renovation. Unlike t-test, regression has abundance variations, I am not even sure how many, but here are three(four) useful regression types for preliminary analysis:
– Univariate linear regression: when you have one independent variable (variable on the right hand side) and one continuous dependent variable (variable on the left hand side), correlate linearly, i.e. bread_sales = constant + discount_constant * discount; the effect of discount to bread_sales is the discount_constant
– Multivariate linear regression: when you have one continuous dependent variable (DV) and many independent variables (IV), i.e. bread sales as dependent and discount, gender, age as independent variables. You can also do interaction here i.e. to answer question like: does bread sales increase or decrease in the event of discount and is it female who drives the effect? so your interaction will be on discount and gender. The information you will get is, does discount correlates to bread sales and if this correlation only happen to female buyers or not.
– Logistic (logit) and Probit regression: If your dependent variable is binary (take value of 1 or 0 only; discrete), also have univariate and multivariate.
IMPORTANT: You have to make sure that the regression equation you are using (basically the formula DV = c + IV_c * IV) is correct. Correct in a sense that there is no problem of heteroskedasticity, multi-collinearity and endogeneity. All of them do not sound English and here is my attempt to explain in simple form (and to my understanding, feel free to double check):
– Heteroskedasticity means your error term is correlated with the DV, meaning you are omitting information, in other words there is other important variable that you have not include in the model. i.e. bread price as DV and flour price only as IV, but the bread pricing is actually very much related to labor wage (for example) so you are omitting this variable thus your data suffered heteroskedasticity problem.
– Multi-collinearity, normally you will hope that your treatment in IV will drive the change in DV or no change but changing DV should not affect IV. But if multi-collinearity happened it means that if you change DV, IV also change. So it becomes confusing because when you change IV, DV is expected to react but reaction in DV caused change in IV, it becomes never ending loop and you realized that both variables are dependent to one another. i.e. bread price and bread supply, if bread supply is high they reduce the price but when price is low they need to load more supply because more people buy them so supply will increase and loop.
– Endogeneity happens when at least one IV is not independent in a sense that it is a decision made by the subject and it is directly affecting the outcome of DV. i.e. bread sales as DV, discount as IV but you can only get discount if you ask for discount, so this become a filter, only people who are sensitive to discount will ask for a discount or something like that.

Still about regression, but lastly, how to know if your model is good? Look at the R squared and F test. Higher R sqaured is in general better but it could be misleading. Especially for marketing data such as survey, you might not want to have too high of R sqaured because it might indicates multi-collinearity instead. Because it is very hard to have such a good model in a real social situation where so many factors are moving simultaneously (you cannot really control everything). But if your data is a hard data or lab data which involves very little human or have extensive controlled environment then higher is better. Just for reference, a professor told me for marketing survey 0.2 – 0.3 R squared is considered good while for biology lab data at least 0.8 is needed. Read more about the topic to get the wisdom to say, is this R-squared real or due to problem. By the way R-squared is a metric which measure the fitness of your data to the generated model, so they look at how much deviation/variation by comparing the observed data to the model defined.

~to be continued


Coding Might be the Last Thing You Want to Do First – Step by Step (Part I)

As promised, this is a continuation of why coding might be the last thing you want to do first. Previous post talked about the disclaimers, suggestions and starter kit, make sure you read at least the first part of it.

Not less importantly, make sure NEVER alter the data, report a made up results, basically don’t undermine the importance of ethical approach, if you found nothing the nothing is your finding, DON’T discount your ethics only to get an A in your course, it is a crime and a disgrace to yourself and your work.

Retrieved from Google 3 May 2018. All rights reserved.

If all is clear, let’s dive directly to the how to data analyzing better:

Step 1: Do you have enough data?
Check if you have enough data entries. This is called the power test, the idea is to see how much error can you tolerate with N sample size. Some people did the calculation beforehand to make sure their results will be robust. The bottom-line is, more is better. It is hard to say precisely how many data you should have but if you do truly random sampling, collect at least 40 for each group, this will allows trimming error entries and satisfied sample size requirement of 30 for central limit theorem to work. But again, more is better. If you are not doing random sampling, i.e. you asked your friends only then you will need much more, and there will be network effect, etc. Not ideal but you can mitigate this problem by asking diverse people in your network portfolio such that they can spread the survey to more widespread type of people (see snowball sampling). If you are using empirical data, you should be able to simply analyze the whole dataset.

Step 2: Trim, feel and clean the data
Trim the data, exclude all non-subject entries, i.e. the sample entry you inputted when you tried if the survey works or not. Also trim the unfinished survey.
Feel means descriptive statistics. After you get your data, the best way to understand your dataset is to run a descriptive statistics. This will give you an overview of variable list, format of each variables, missing values, mean and standard deviation (SD). I would first look at missing values and the sample size (N) of each variable. If each variable has different N then we are missing data, what does this missing value mean? no applicable? or is it a “No” answer? Then I will look at the mean and SD, for example I know that most of people who took my survey was female (coded 1 and male coded 2) but the mean is 1.8 or 0.9, those are weird.
After you felt the data and spotted weird events (you hope you don’t have these but they will be there), you need to clean them. And it is a tedious work especially if you have large dataset. I would suggest first sort the format, make sure all numerical be numerical, ordinal marked ordinal, and string as string. Then, fix those weird things you found from feeling, for example re-coding the variable, making sure that answers are logical, i.e. if you asked for age and someone said 4 then that is not logical. But if they said ‘4q’ then maybe it is a typo of 41 or 42, use your wisdom to choose. After those are sorted, you might want to do is to grab a pen and paper, jot all variables you have, identify questions which asking similar question, are they reverse coded or not, then pair/group them. In this process you should pay attention to logic related variables, i.e. if someone claimed to be 20 years old but said that he has speak English for 21 years then it is not logical.

Step 3: Go back to basic – Detailed the question, find the variables
After you are sure that your data is clean and pretty (maybe not so pretty but neat), move on from it, leave it alone but the list of variables within your heart. Forget your data and go back to your research question. Now if you have a question like: Do female react more to discount than Male?, good but not enough. Define what does react more mean? Buy more? come more? more willing to buy? or? Then determine the measure from your list of variables, which variable(s) indicates “react more” as you defined it. Realized from that question, is it also unclear what discount? how much? For example, if the discounted good is bread then you might want to see the change in bread sales and not the cheese though they might related. Also, the effect of 5% discount and 10% discount are different, especially for the more expensive goods. Which condition are you observing? try to remember your research design, and write as detailed questions as you can.
And one question might not be enough. Of course you should and recommended to have one big research question, but when want to understand it, don’t just look for one story because, one, it might not work, you might not be able to find your answer, i.e. you just cannot find out if female consumer buy more bread than male when there is 5% discount due to lack of data in gender information in the transaction data/survey (I know it looks like a stupid example, but this does happen). Two, it might not be the right answer as you are missing another important side story, i.e. female buy more bread than male when there is 5% discount but maybe the female in your data are buying groceries for the whole family while the males are just buying for their breakfast, so it seems that female is buying more when it is really just a conditional reason. Three, you might find something more interesting and beyond your research plan. If you restrict yourself to looking only to what you think you will (or will not) find about one thing, you might be missing out a big story inside your dataset.
So, ask more questions and ask with greater details. Practically you do this by writing all questions and hand pick variables you think will help you answering these question. Analogically speaking, it’s like you found a good dish in the internet but no recipe available, you went to buy the ingredients you think you would need according to your experience (literature review) and now you are ready to figure out how you will do it.

~to be continued


Coding Might be the Last Thing You Want to Do First – The story

Weird title but if I misdirected you and made you expect to read reasons why you should not code, then feel free to close the tab and move on with your life. Why? Because what I want to cover here is how to do data analysis which involve statistical coding and why the coding part should never be the first. This is the first part and will tell you the disclaimers, the motivation and the starter kit. For the step by step, go to the following post (will be ready before May 10 2018), thank you~

OK, here are some disclaimers:
First disclaimer, I am no professional in the field. I do data analysis as part of my work but I feel there are a lot more that I can learn in the field so I am not a novice but far from professional.
Second disclaimer, this is just an outline of what I think is the better approach to handle data but there is always the best way to do data analysis uniquely to your dataset, so this is not a magical procedures and might not work in couple situations (though in general they should work just alright).
Lastly, this is preliminary, so if you have basic in data analytics, marketing research, data handling, don’t worry if you miss this post.

Retrieved from Google on 3 May 2018. All rights reserved.

I am writing this because of the recent event happened around me. In academic circle, May is a very tense month because this is namely the month when projects are due, the ending of classes (meaning the hardest topic for most of the course were just covered, or maybe intense because the lecturer hasn’t finished covering all materials) and final exams. Out of a whim, I offered my help to a group of distressed UG student group doing marketing research. They had some survey data and need to make sense of it with SPSS, which I have basic understanding in and maybe because I have some free time on May 1st, I decided that I should lend a hand (if they still need the help and they do). I was so nervous because I am afraid that I gave them “too much hope” and what if I failed to give what they need, or worse, if I make them more confused because of my own confusion, but anyway, I met them. Long story short, I managed to keep my calm and I think I helped them. Then somehow another team also sought my help and (I hope) I helped them too.

What fascinating to me is, the first thing they asked were how to do the SPSS. How to clean, to do the regression, how to do the t-test so on and forth. I was like “Woa woa, take a step back, what do you want to do?” So they told me the regression and tests they want to do but when asked why they want to do those they have very vague idea why. I think they see it as ok do the regression, t-test, make it look fancy and write whatever results it gave. Statistical tools seem to be a black box in which one input data, click here and there and it will give results that hopefully will answer the research questions. But it is not. So before you go to the step by step post, here are some starter kit which might help you in doing your data analysis

  1. Choosing survey platform (if needed)
    Assuming you know how to write proper survey question, choosing and understanding what your survey platform will give you is a beneficial thing to make your data cleaning (most tedious process ever) and your life easier. What you might want is to be able to restrict they answer format, i.e. for age question, restrict only integer number.
    And find a platform where you can get the data in most convenient format. i.e. if you have tens of thousands data entry with double digit variables, you might don’t want them in .xlsx form as it will take forever to download and open. In this case .csv might be better or directly download in form that readable by your statistical tool. But if you have small data, excel might be enough.
  2. Choosing statistical tool
    Similar to the format mentioned in point (1), the way to choose your statistical tools also related to the shape of your data and research purpose. If you only want to do something preliminary with small dataset, go for excel, it is quite powerful and FYI, you can do simple regression there too!
    For larger dataset, you can opt for R, SPSS, Stata, Minitab, SAS, etc. I believe only R is free so that’s the first consideration, but if you are not restricted by the price, then ask yourself are you sophisticated in coding? If no, go for SPSS. This is the most simple (I found so far) drag and click statistical software. If you are a little more into codes go for Minitab or Stata, here you can still drag and click but if you understand the code, it makes your analysis more sophisticated (easier and faster too). If you are good at coding then go for R while if you are proficient go for SAS.
    Pay attention that all each package has their own language, some might overlap but not so much. But the logic is all the same (except for SAS and excel) so normally if you can do one thing in one platform, just google it and you will find the command ‘translation’ to the other platform (except for excel, and much more function available for SAS).
  3. Always save your code command
    If you are using Minitab, Stata, R or SAS or any statistical tools that need a degree of coding, always always always save your codes. This will save you in case you screwed up somewhere and need to trace back on what you did wrong. If you are doing it in SPSS, detailed the steps simply in Word/notepad. Feels like a pain but it is like an insurance in case in the future your data encounter a bad fate.
  4. Just a suggestion 1: make a new variable for everything, never overwrite the original data especially if you are not sure if you can get the original data anymore.
  5. Just a suggestion 2: You wont need to keep saving your dataset with each change (that will give you xxxV1.dta, xxx V2.dta, … , xxxV10-(final).dta), just save your code file regularly. If something happen you just need to re run that file and voila~ If you found anything wrong then go to that file and fix the code, re run and voila~ — You get what I mean.

So I guess now you should be ready to read the next post (will be ready before May 10 2018). See you!

Labour Day: About Job Equality

So, the hottest trend award this year seems to go to #MeToo movement and all kinds of women-led movement to promote gender equality. I would not dare to judge these movements as good or bad but I think I have about enough dose of news on them everyday and now it’s time for me to share my own thought on it.

Retrieved from Google on 1 May 2018. All rights reserved.

Surely it has been widely known that in the past, women were expected to stay home, take care of the husband and children. Later, the change started to happen in the West area and women were then allowed to work. But the type of jobs were really limited, maybe nurse, cleaners, or secretary, but never high stake occupation. Much later the trend again changed in favor of women, made them able to actually have better education and thus opened the window of opportunity to take position in the higher ups. Now, as we can observe, much more females are working in important positions, there are even some who successfully ran for presidency, central bank chair, etc. The recent hectic are trying to either open the opportunity windows to every women or to debunk the hardship that women had to endured in their position, to stay there or to climb the managerial ladder. I sincerely hope that this will lead to the best result for the society.

But there is really more to it than just female shouting for more rights. I was inspired by a speech by Anne-Marie Slaughter. I believe that we have been looking at the world in a distorted way. The discrimination happened not only because female is viewed as inferior but because the job female does are generally labelled as easy job. We know and have to admit that on average female are different from male, but each individuals are different from one another too, with each has their own strengths and weaknesses. Using the economic term of comparative advantage theory, we should let people who has the highest comparative advantage to take the position. So if one person are better at cleaning than staring at excel template crunching number so he/she would be better off doing the prior. What we are missing is equality in looking at the job title. It seems like in general people think that being a doctor is more important than maybe a maid at home, clear reason because doctor can save your life. But I invite you to think it this way, your maid might also be saving your life a big time. Say, maybe without you knowing, your maid had helped you to turn off the stove you forgot to turn off, threw away leftover food which otherwise will stay in your fridge for only God knows how long, making sure the house is clean and free of harmful bacteria, not all because it is not possible but at least partially. You are scared to piss your doctor off because he might kill you without you knowing, but the opportunity that your maid had to kill you is higher than the doctor. And they can get away with it too if they are clever enough. So why are we underestimating the work of maid in comparison to doctor?

One possible reason is maybe monetary compensation received by some workers and far less than the other. But are those ‘high-class’ job really provide better life quality than in the ‘lower-class’ one? In this increasingly materialistic world I would argue, yes! having more money most likely will lead one to have a higher life satisfaction simply because you need a certain amount of money to life comfortably (study). Now consider two jobs, butler (male for maid) and accountant in the United Kingdom. According to totaljobs, in 2018, the average pay for butler is around 32,500 pound (small sample size of 24) while 37,500 pound for accountant (large sample size of 1,331). Well accountant still pay higher than butler but if we take into account the housing and food, maybe net income of accountant might be lower than the butler’s. But still, the job accountant is perceived as more honorable and desirable, they think that people who work as an accountant can easily do the work of butler but not vice versa–I doubt so, with enough training, one can learn almost anything, they might just not be as good as the other person or simply has other things where they can strive in.

Retrieved from Google on 1 May 2018. All rights reserved.

So I think I have illustrate the discrimination in evaluating different occupation. It turns out that job with lower ‘caste’, are job which female are the majority in the workforce. We need time machine to study does which caused which — namely was low level job category founded first such that no man wants to take the job and women who entered the market late has no choice to do these jobs or because the job workforce were mainly woman initially such that the job gained their label as low level job — cause a job has an implicit (sometimes even explicit) label as an easy job. If we can agree that all jobs are equally important, then there should not be problem of which gender do which job, all jobs are equally respectable. Each job has their own importance, some seemingly have more weight than other just because, resembling human, every each of them has their own role that concerning different life. You cannot compare the importance of one thing to the other when they are different by nature, president might be very important occupation because he/she make decision that concerns everyone in their country and maybe other country’s residents as well. But if all farmers are now stop being farmers and run for presidency or work as anything else, then no one can survive because there is no food available anymore due to absence of farming job.

Retrieved from Google on 1 May 2018. All rights reserved.

For the closing remark I would like to underline that above, I argued that all jobs are equally respectable, I am not trying to imply that every job has to pay the same or anything with the same sounding as that. Salaries shall be proportional to the amount of work, incorporating risk within it. Long hours and good quality work should be appreciated. So, every job with different position and responsibility should pay different level of wage, but NO MATTER what is your occupation and how high/low your position is, never underestimate others and respect everyone. Lastly however, I really think that CEO paychecks are far too high especially in big multinational companies i.e. Walmart. Thank you for reading, happy labour day 🙂

Quarter Life Crisis and Happiness

If you are someone in undergraduate study or below, this might not be very applicable. But for those who had just graduated, raised your hands up if you feel like you are lost in life; I welcome you to the QUARTER LIFE CRISIS club.

Retrieved from Google on 7 April 2018. All rights reserved.

In the past weeks, I tried to catch-up with various friends from my university time, I realized that I had missed out a big time. Some had changed their occupations once or twice, some other had gotten themselves yet another ex-boyfriends and boyfriends, the younger ones are going to graduate very soon and preparing for the coming job, etc. We talked about how our life went and where will it go from there on. Only few were optimist about the future, while the majority felt uncertainty and skepticism when asked about the near future. For me, I told them, I live in this moment, currently I feel pretty comfortable with my life and I have a decent short-term plan, though I am far from optimist about the future. So, one typical question was asked: “are you happy with your current life?” My answer was, happiness is over-rated. Why is that? well, research has shown that people who try to find happiness in life, in the end of the day tend to be less happy. Because, what is happiness? How to be happy? Are you happy if you are the only one who is happy? How do you measure happiness and is that the ultimate goal in our life?
Surely, no one wants to feel sad, or depressed or have a burdening life. Everyone in general will prefer to have a light, lovely and bright life. But how to achieve that? Simply by having fun all day? Smile, laugh and be happy? In my opinion, it is not that superficial. Life is more about your purpose, like the ted talk by Emily, we need an objective to feel fulfilled in life. To achieve that, we might need to go through all the bitter and sweet events in life, but ultimately, we will reach the state of joyfulness.

I am now in the middle of finding the right path to seize my goal flag. I feel overwhelmed. I feel like this is a never ending rode that I have to go through but only uncertainty awaits me in the end of the rode, if there is an end to this. So even after I found my purpose, it does not necessarily mean that the effort stops there. It was just the starting point, every subsequent steps will always be harder, tougher and sometimes it will force us to bend down. The only remedy might only be the thoughts of achieving your passion.
But then I realized, when I look back, I saw familiar faces, loving faces who supports me, and I know they will be there for me in good and bad times. So, even though I must walk this path with my own strength but having people who walk with me can be another source of strengths too.

Life is full of different colors, that made it beautiful. Original picture by @random.walk_. All rights reserved.

In the face of quarter life crisis, just take a deep breath, take a step back and look at things from bigger perspective. I know I might not be able to understand wholly every one of your concerns, but there are people around you who want to open their arms to provide their time and ears. They might not give the best solution but it is always nice to know that you have someone. It is ok to rely on others every once in a while and do also provide the assistance needed by your other loved ones.

Recommended Ted talk:
Emily Esfahani Smith – There’s more to life than being happy
– Sophie Andrews – The best way to help is often just to listen