5.2 Content Analysis: Quantitative Research – Observation and Content Analysis

behavior is not the only thing that can be observed we can also observed and count documents and this brings me to the last element of quantitative data collection that i'd like to discuss which is content

analysis analysis is an approach to the structured analysis of documents and text and content analysis deals with quantifying content

in terms of predetermined categories so it is very similar to structured observation in the sense that it tries to put something more complex behavior on one side text on the other side into a structured format by using

categories it's systematic and reliable as any of the quantitative techniques that we discussed and this makes it different from related methods semioffice or ethnographic content

analysis which we will discuss later when we talk about qualitative research studies in many many cases content analysis will relate to the analysis of mass media usually print but it can also be

other things and in this case the question of sampling comes up as usual and there are two possible dimensions here the first question is which media do you want to sample which type

is it only printed only or also visual data documents mass media whatever obviously print text is less rich than pictures or even videos but

in exchange videos and pictures are harder to quote which creates methodological challenges if you decide on using on mass media which kind tv

radio newspaper magazines more than one type and again how do you make the results comparable there's more than one type and for each type of text which type of examples which type of newspaper

because there are many different types different countries different languages and so on and so forth all of this needs to be specified and defined as precisely as possible

to ensure replicability later which is one of the key cornerstones of most quantitative research methods and once you have decided the media the question comes which dates

do you want to sample the starting dates may be predetermined by some kind of historical events there may have been something happening i don't know the subprime mortgage crisis the outbreak of corona

and you may want to understand how this event impacted things that happened afterwards in some cases the date may be more open if you are looking at an ongoing general phenomenon

if you want to look at the presence of women in firm leadership there is not a specific date that you get used but you may use a time frame that makes most sense to you the same about the end date which is

very often a matter of judgment because at some point you have to stop collecting data and however when you select the media and when you sample dates keep in mind the general idea behind

probability sampling we have already discussed this we talked more about this next section because these should guide most of your choices

now how to do content analysis the first question is what do you want to count and here are some options you may count the actors that appear the names the protagonists and alternative voices

you may count words that are used or even entire phrases you may try to count and categorize subjects and themes any type of disposition and value judgment bias ideology

you may want to look at pictures and specifically how they look at like when you look at marketing for example where pictures can play a very important role there's a question of the coding schedule which similar to anything else

that we discussed is a tabular form in which you enter data usually each column represents one dimension and each row represents one unit of text and

then you fill in the information and there needs to be a coding manual because as always it may not only be you who codes this you may have research assistants you may have people helping you with this so there need to be a set

of instructions which needs to list all particular categories in each dimension show which numbers and codes refer to which category gives guidance probably using some

examples explains what to do if more than one category could imply and because there is a lot of work involved in this many in many cases you would simply use

some other guide that has been done for another study and adapted to your purpose there are a couple of pitfalls that you have in content analysis one is the definition of the research

question it must be very clearly specified before the analysis content analysis is a quantitative method so you cannot adapt the research question on the way you need to define up front what you're

looking for up front you need to decide which dimensions of the text to cause if to clarify and to to record you need to decide how many different views you want to represent

what is selected for inclusion we talked about the topic of sampling what do you decide to take out what if the amount of coverage changes over time how do you deal with this

you need to ensure that your coding scheme is very easy to understand it needs to have discrete dimension as always mutually exclusive categories exhaustive categories clear instruction

and as in all the other cases usually a pilot study makes most sense to ensure that things are consistent between coders and for the same coder over time

now when does content analysis make particular sense well the advantage is the approach is very transparent because you have defined everything in principle your approach could be replicated by anyone who simply follows

the same rules and procedures that you have followed it allows for a very longitudinal analysis some newspapers go back decades probably even hundreds of years it's unintrusive

there is no reactivity bias that we have in observation because the news have been reported independently of how you look at them a decade later it's flexible and in most cases the information is very easy to access

because nowadays you can access it online the disadvantage is there's always a question of authenticity there's always possibly a bias certain newspapers may report things

differently certain journals may report things differently in certain points in time things might have been reported differently in many cases it it depends on impression of coders

which you can avoid by having a clear schedule but which always remains it may make invalid inferences of the content you may read the document differently

today than it was intended to be read 20 years ago as behavioral analysis content analysis cannot answer the why

it can only answer what happened and it is purely a theoretical it means you look at the content and you try to aggregate and make sense of it

and with this we end the methods of quantitative data collection in the next section i want to speak a little bit more about the question of sampling which i approached already a couple of times before

and finally how quantitative data can be analyzed in the most efficient way