You can test various hypotheses using content analysis. However, it is important that you formulate these hypotheses clearly and concretely prior to performing the content analysis. Using content analysis, you can extrapolate your findings (i.e. generalization them to a broader context, which you have not actually measured). Examples of such extrapolations are trends, patterns and differences. From these trends, patterns and differences you observe in the collection of documents that you analyzed (your sample), you can draw conclusions about similar documents that you have not actually analyzed (the population) and thus test your hypotheses.

Trends

A trend is an association between time and a measurement. A trend hypothesis argues that this measurement depends on time. The basic hypothesis (H0) states that the average value of the measurement remains stable over time or only changes 'by chance'. AmCAT provides you with various possibilities to analyse trends. Using AmCAT, you can visualize a trend graphically any draw preliminary conclusions about the trend. You could, for example, test whether a trend is linear or cyclic. To display a possible trend in a graph you can use the graph function in the Graph/Table function. AmCAT also provides you with the possibility to produce data with which you can perform a formal test of the trend hypothesis.

As an example we will use an article set called 'Wikinews articles for terror*', an article set with all Wikinews articles including the word 'terror' from November 2004 until January 2015. An example of a trend hypothesis could be: "The attention for Isreal in the Wikinews articles on terror* has decreased in the period from 2009 until 2014". Figure 6.5.1 diplays the attention for Isreal in Wikinews articles on terror* per year from 2004 until 2015. This graph was made with the Graph/Table function, using the search term 'israel*'. The line in the graph suggests that, overall, the amount of attention for Israel has indeed decreased, but this may also be the result of natural fluctuations. To be able to really support our hypothesis that the attention for Israel in the Wikinews articles on terror* has decreased in the period from 2009 until 2014, we need to perform a formal statistical test. To do so, we first produce a table with the measurement of interest (attention for Israel) per time unit (year) by selecting 'Graph' as the output option in the Graph/Table function. To analyse the data in a statistical program, such as SPSS, we export the data from the graph by clicking 'Export Aggregation'. Next, we open the resulting CSV, Excel, SPSS or HTML file in a statistical program and perform the analysis of interest (e.g., a correlation analysis). The result of this analysis will enable us to conclude whether or not our hypothesis is supported.

6.5.1. - AmCAT Navigator 3 Testing Hypotheses Trends

Patterns

Differences

AmCAT Version
This page describes a feature in AmCAT
View other version: 3.3 - 3.4 - 3.5