As the question states, I'm interested in the difference between CIs, SEM, and SD, explained in plain words (so as to be able to communicate it to people without maths/stats background).

My current understanding (mainly gained through Googling around) is that:

  • CIs indicate the reliability of a measurement
  • SEM reflects the uncertainty in the (sample?) mean and its dependency on the sample size
  • SD informs us about the spread of the population (i.e. SD reflects the variation of the data)

Is that correct?

  • 1
    $\begingroup$ See stats.stackexchange.com/questions/32318/… and stats.stackexchange.com/questions/6652/… , you can also easily search the site for multiple similar threads. $\endgroup$
    – Tim
    Apr 8 '16 at 8:29
  • $\begingroup$ Thank you for your input, Tim. However, none of the questions/answers you've linked to answer my question, which is about a plain explanation of the concepts (without assuming any prior knowledge of mathematics or statistics). I also, of course, search for similar questions before I posted mine, but I couldn't find the answer I was looking for. $\endgroup$ Apr 8 '16 at 8:40
  • 1
    $\begingroup$ I have edited the title to more closely indicate the intent of the question. A good answer to this question could easily have served as an answer to one of the existing questions, but I do think the "plain words" aspect might serve to differentiate it and I can imagine answers to the other questions being confusing to non-specialists. I also wonder if bundling CI, SD and SEM together makes it too broad, but on balance I think they are sufficiently closely related to make a combined question acceptable in scope. I'm, just, inclined to vote to leave this open. $\endgroup$
    – Silverfish
    Apr 8 '16 at 9:17
  • $\begingroup$ CI does not imply reliability of measurement.Pl. understand statistics is different from psychometry. SEM attempts to assess uncertainty and not in a particular sample mean. SD tells spread of sample mean and not about population. $\endgroup$ Apr 8 '16 at 9:31
  • 1
    $\begingroup$ @subhash I think "SD tells spread of sample mean" needs some clarification: it measures the spread of the data about the mean, rather than the spread of the mean itself. $\endgroup$
    – Silverfish
    Apr 8 '16 at 9:36