Many of the international datasets offer a facility to analyse the data online. For straightforward descriptive analysis or for initial exploration or assessment of the data this is probably sufficient for your needs. For more complex analysis you will want to download the microdata for analysis with a statistical software package. While it is possible to analyse many datasets online using It is often the case that the dataset does not contain the variables you wish to use in the right format. For example, you may be interested in a particular cohort but the variable on age is in 5 years bands. You need to create a new age variable with the bands of interest for your analysis. Similarly, you may intend to run a multivariate analysis where you need to set a reference group or create dummy variables so expect to spend some time getting the data in the format necessary for the type of analysis you want to do. For example, you may wish to create a dichotomous variable as to whether someone is religious or not from a question or set of questions on church/place of worship attendance, self categorisation of religiosity or frequency of private prayer. As the norms of collective worship or frequency of prayer differ between religions you will want to try and determine a best fit definition. As the distribution of affiliation to particular religions varies been countries you will need to be sure that where you identify differences between countries you are not simply finding differences between religions and vice versa.
Download and study the ESS Round 2 Gender weighting spreadsheet
These tables show the effect of weighting on the gender composition of the sample and illustrate what the population weight does. This reduces the contribution of the sample for any given country pro rata to its size.
Download and study the spreadsheet on interview length.
These tables show the effect of weighting on interview length - the first table is unweighted, the second has design weight and the third has the design and population weight.
Before conducting any analysis stop and ask yourself if the variable meets the requirements for that analysis technique. For example, does the analysis assume that the data is normally distributed? Does the data need to be continuous? Is there a requirement of independence?