Statistics needed to understand data?

6 10 2011

After going through a lot of hard work coming up with hypotheses, going out and carrying out experiments and research, to get your results and see what’s going on makes it all worth it. But do we need statistics to understand what our data points to?

Well, the simple answer is: yes. Without statistics, our hard-earned data will be completely meaningless; just a bunch of numbers. The only time you probably don’t need statistics to see what’s going on is if you have a very small sample, but that’s no good for generalizing at all.

Even simple statistics like mean, median, and mode can give us a great insight into what’s going on with our data. Mean seems to be the best bet out of these four, but it can be influenced by extreme outlying figures. The median isn’t influenced by these extreme figures, but it doesn’t really tell us much about the data. Also, the mode can be pretty misleading, for example, you could have a sample of 100 people, maybe even 1000, and they all have different scores, except 2 people who share two EXTREME scores.

If you’re doing some serious research and have a rather substantial number of participants, the  amount of data you’ve collected at the end can be pretty overwhelming. This is where statistics can be life savers. If you’ve got a program such as SPSS at your fingertips then in a matter of seconds you can find the mean, median, mode, range, variance, inter quartile ranges and so on. You can also find the p value, which can tell you whether you’ve found a significant difference between two conditions or not, meaning you can either prove or falsify your hypothesis.

In conclusion, I think it’s pretty clear that we need statistics to understand our data, you’d have to be really passionately and stubbornly against statistics not to.




15 responses

14 10 2011
14 10 2011

I agree with your view on the question and I liked how you talked about the validity of your results in relation to statistical analysis. However I would like to argue that if you are not familiar with the use of such programs such as SPSS you can still make invalid inferences from the results due to having issues with using the software as I am sure many of us discovered at one point or another last year. 🙂

14 10 2011

I enjoyed reading your blog as it was concise and to the point and I share most if not all your views. However, I felt your argument displayed how much we need SPSS to understand our data rather than statistics. That being said I will definitely be reading next week’s post 🙂

14 10 2011
14 10 2011

I agree with your points in the blog, statistics are needed to gain control over enormous amounts of data. However, applying your points about outliers into this information, would the statistics really help? I think it would be necessary to scour through data before applying statistics to find averages on a large data set that may have a couple of extremely out of line data points. “…outliers are often bad data points.”

14 10 2011

I do agree with your point but I would like to ask whether you think that statistics is unnecessarily overcomplicated? As we have all seen an SPSS output and there is so much information on it that proves very little use (except when being asked in a stats test) as we only use the significance when reporting our results – as that is all we actually use in any investigation. I’m not saying that for all data we should just compare the Mean, Median and Mode but there is a lot of data that no matter how complicated the stats are. Especially as the they will never account for Individual Differences no matter how much extra stuff gets crammed into the output

14 10 2011
Homework for my TA – Week 3 « My Statistical Blog

[…] 7) […]

14 10 2011

Yes statistics is necessary and I’ve mentioned this on other blogs, but you don’t seem to have considered qualitative data as a viable source of information.

Not all data has to be numerical, some is verbal, observational or prose (like questionnaire answers, after all opinions can also be data) and so whilst it might not be necessary to use statistical analysis qualitative information still counts as data. (

14 10 2011
14 10 2011

I would suggest that not only do you need statistics to understand your data, you need a good understanding of statistics in the first place, otherwise you won’t remove the outliers which can dramatically change the data ( or the drop out data or the data where the participants were obviously just not paying attention. Statistics will not remove those for you or tell you if there are outliers and therefore you need a good understanding of statistics to select the right data to be analysed.

13 10 2011
Homework for my TA « prpsjj

[…] Share this:TwitterFacebookLike this:LikeBe the first to like this post. […]

13 10 2011

I liked how you talked about the mean, mode, range etc because I think that people often forget about these simpler statistical methods believing that the only statistical method there is is t tests/p values/significance and then jump to the conclusion that we do not always need statistics because t tests are too insensitive or miss things in a small sample. Where as i feel that t test are not the end all of stats and that when they are not appropriate we can use another stats method such as the ones you mentioned or a graph of some sort. However you say that the only time you don’t need statistics when you have a small sample, but maybe you do need statistics, because even if the sample is as small as say 4 you may still gain further knowledge by using one of the many statistical methods.

13 10 2011

I totally agree with what you have said! Statistics is a way of adding meaning to your data and enables you to interpret it. To prove or disprove a hypothesis you would need a lot more that just a bunch of numbers. Without statistics you wouldn’t think to look for outliers and extract them from the data…this could then cause the data to be misleading and misinterpretated. So yes statistics is needed to analyse data else you are limited to how far you can take your analysis and after the effort of collecting it you would think you would want to go the extra step! 🙂

13 10 2011

It’s true that stats are very useful for cutting down our data into manageable numbers rather than having to trawl through hundreds or thousands of individual scores. You mention in your blog that having a small sample is no good for generalising, but surely that depends what you would consider a ‘small’ sample, and how big the population is that you are trying to generalise to? For example, if your target population was a school year group, of around 200 students, then to get a representative sample size would take far less people than a representative sample to represent all the people in the U.K.
100 participants from the school would definitely be considered representative, but 100 participants to represent the whole of the U.K. would not be representative at all.

7 10 2011

Totaly agree, statisics is majorly usefull when summeriesing data and seeing how sinificant the data is. I would like to stess (just to fill the wordcount, clear and presice blog by the way) that stats such as p or N isn’t well known to the genral pop ( based around people i know and myself before last year) . This means that although stats is needed to understand dataits not acessible to readers without a understanding of stats.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: