Table 3 Key measures that are vital for interpreting the results.

%_wtn_mth (Percentage within month)

In relation to recovery activity, %_wtn_mth is the percentage of the total recovery-related tweets for any given month that relates to a specific recovery activity. This measure is important because it tells the extent to which each category of recovery activity is represented in the total recovery-related tweets generated for any given month

Whereas, in relation to user group, %_wtn_mth is the percentage of the total recovery-related tweets for any given month that is generated by a specific user group. Similarly, this measure tells the extent to which each user group contributed to the total recovery-related tweets generated for any given month

%_wtn_group (Percentage within user group)

%_wtn_group is the percentage of tweets from a specific user group that is generated in a given month. This measure is vital because it tells how the total tweets from a specific user group is distributed across all the months investigated

%_wtn_rec (Percentage within recovery category)

%_wtn_rec is the percentage of tweets for a specific recovery activity that is generated in a given month. This measure is vital because it tells how the total tweets from a specific recovery activity is distributed across all the months investigated

Note %_wtn_mth, %_wtn_group, and %_wtn_rec are relatable to the statistical concepts of 'percentage within column’ and ‘percentage within row’ that are commonly used in contingency tables³⁷

Standardised residual

\(Standardised\, residual=\frac{Observed-Expected}{\sqrt{Expected}}\) (See³⁷ and³⁸)

This measure shows the strength of the difference between the observed and expected values

We have used contingency tables to record and analyse the joint distribution of any two variables. For each contingency table, the count in a particular cell, \({x}_{ij}\) is the value of a random variable from N samples with a multinomial distribution³⁹. Suppose that \({X}_{i}^{^{\prime}}\) represent the sum of counts in all cells along the i^th row, and \({X}_{j}^{^{\prime}}\) represents the sum of the elements in all cells along the j^th column. If the two variables involved in the Chi square test are independent based on the null hypothesis, then the expected value of the random variable \({x}_{ij}\) can be calculated using the formula³⁹

\(E\left({X}_{ij}\right)= \frac{{X}_{i}^{^{\prime}} {X}_{j}^{^{\prime}}}{N}\)