It appears this study shows that the majority of suspects for crimes are immigrants, which is an important distinction. I have no knowledge about the situation in Sweden, but in Germany there are known statistical problems like immigrants both having a higher chance of becoming suspects ("Tatverdachteffekt") and crimes where an immigrant is suspected are more likely to be reported ("Anzeigeeffekt").
Stastistics about suspects are most commonly used in studies like these because the police, due to the seperation of powers, usually has no or at least less statistics of the actual results of charges.
That low resolution rate alone introduces a margin of error that is larger than the difference between immigrant and non-immigrant suspects. It is therefore possible that the much higher immigrant suspect rate is entirely a result of biases.
Stastistics about suspects are most commonly used in studies like these because the police, due to the seperation of powers, usually has no or at least less statistics of the actual results of charges.